The goal of this research is to develop, test and apply a generalized planning method which allows one to approach a set of data describing patient attributes, specify some set of dependent demand variables (such as length of stay, post-operative length of stay, units of ancillary services such as x-ray and laboratory services) and examine logical structures of attributes which from a medical point of view would give a statistically consistent prediction of the dependent variables. Then resource allocation models can be constructed, depending upon the mix of the diagnostic and social characteristics of the patients to be treated and the programs and facilities planned, to meet these demands. Our approach is expected to result in a complete planning system consisting of the following components: 1. A patient classification system to allow for the definition of homogeneous classes from the point of view of resource consumption. 2. A patient path generator which will associate patient classes and their resource consumption with specific facilities to be used. 3. A path simulator which will predict the resources consumed by any mix of patient classes. 4. A resource allocation model which will translate resources consumption into budgetary goal attainment, and service level terms to that more effective planning decisions can be made.